Reported Speech in Swedish and Ten Immigrant

Projektet SVENSKA SOM MALSPRAK ( SSM )
The project SWEDISH AS A TARGET LANGUAGE
SSM REP ORT 5
Bjorn Hammarberg and Ake Viberg
REP ORTED SPEECH IN SWEDISH AND TEN IMMIGRANT LANGUAGES
Stockholms universitet
�nstitutionen for lingvistik
November 1976
Rapportserien SSM Reports
SSM-rapporterna (serien SSM Reports) utgavs vid Institutionen för lingvistik, Stockholms universitet i anslutning
till forskningsprojektet Svenska som målspråk (SSM) som understöddes av Skolöverstyrelsen 1973–1980.
Redaktör för serien var Björn Hammarberg. Projektet var inriktat på studium av sådana företeelser i svenskans
struktur och användning som utgör problem för vuxna inlärare av svenska som andraspråk med olika
modersmål.
Upphovsrätten innehas av författarna. Rapporterna får laddas ner för personligt, icke-kommersiellt bruk och
får citeras enligt gängse vetenskapliga principer. De får fritt användas i kurser inom högskolan under
förutsättning att källan uppges.
Förteckning över SSM-rapporterna
1.
Hammarberg, Björn & Åke Viberg (1977), Felanalys och språktypologi. Orientering om två delstudier i SSMprojektet. 34 s.
Även tryckt i Nysvenska studier 57 (1977) i form av två uppsatser, ”Svenskan i ljuset av invandrares
språkfel” av Björn Hammarberg (s. 60–73) och ”Svenskan i typologiskt perspektiv” av Åke Viberg (s. 74–
85).
2.
Hammarberg, Björn & Åke Viberg (1979), Platshållartvånget, ett syntaktiskt problem i svenskan för
invandrare. Andra, omarbetade upplagan. 67 s.
En engelsk version är tryckt i Studia linguistica 31: 106–163 (1977) under titeln ”The place-holder
constraint, language typology and the teaching of Swedish to immigrants”.
3.
Hammarberg, Björn & Åke Viberg (1976), Anaforiska processer i svenskan i invandrarperspektiv – några
utgångspunkter. 13 s.
Även tryckt i Nysvenska studier 55–56: 213–226 (1976).
4.
Åslund, Jan (1976), Araber skriver svenska – arabisk syntax och arabers syntaxfel i svenskan. 47 s.
5.
Hammarberg, Björn & Åke Viberg (1976), Reported Speech in Swedish and Ten Immigrant Languages. 21 s.
Även tryckt i Papers from the Third Scandinavian Conference of Linguistics, red. Fred Karlsson (1976).
Turku: Academy of Finland, Text Linguistics Research Group. 131–148.
6.
Strömqvist, Sven (1980), En orientering om NP, kasus och prepositioner i ryska. 21 s.
7.
Viberg, Åke (1980), Tre semantiska fält i svenskan och några andra språk: 1. Kognitiva predikat. 2.
Perceptionsverbens semantik. 3. Emotiva predikat. 103 s.
En förkortad version av uppsats 1 är tryckt i Svenskans beskrivning 12, red. S. Fries & C.-C. Elert (1980).
Umeå universitet. Uppsats 2 även i Tvåspråkighet, red. A. Stedje & P. af Trampe (1979). Stockholms
universitet. Uppsats 3 även i Kontrastiv lingvistik och sekundärspråksforskning, red. B. Hammarberg
(1979). Stockholms universitet, Institutionen för lingvistik.
8.
Viberg, Åke (red.) (1983), Studier i kontrastiv lexikologi: Perceptionsverb. Andra, utökade upplagan. 133 s.
9.
Studium av ett invandrarsvenskt språkmaterial (1983). 115 s.
10. Hammarberg, Björn & Åke Viberg (1984), Forskning kring svenska som målspråk. Två forskningsöversikter:
1. Grammatik och ordförråd (Åke Viberg). 2. Fonologi (Björn Hammarberg). 64 s.
Även i Nordens språk som målspråk. Forskning och undervisning, red. K. Hyltenstam & K. Maandi (1984).
Stockholms universitet, Institutionen för lingvistik.
C ONTENTS
Reported Speech in Swedish and Ten Immigrant languages
Sammanfattning: An-faring i svenskan och tio invandrarsprak
Notes
References
•
•
•
·
.
.
.
.
.
.
.
.
.
.
.
·
.
.
.
.
.
.
.
.
·
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Tables
A first illustration of speech reporting
1
Example based on table 1
2.
Structure of direct and indirect quotation clauses in
J.
Swe'dis h
4.
Summary of mechanisms of forming indirect report
•
•
•
.
.
•
•
•
•
.
•
.
•
•
•
.
.
•
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
•
.
.
.
.
.
.
.
•
•
•
•
•
•
•
•
•
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
•
.
•
•
.
.
•
•
•
•
•
.
1
18
19
19
1
2
4
17
- 1-
REPORTED SPEECH IN SWEDISH AND TEN IMMIGRANT LANGUAGES *
One thing that we often have to do in human communica­
tion is to report speech, for example to tell or ask somebody
what somebody else has said. From the point of view of immi­
grant instruction, it is interesting to find out if the means
of expressing reported speech differ in various languages.
Table 1 gives a simple illustration of the speech-report­
ing situationg Initially, we have the situation that is given
in square brackets in section 0, the fact that Eo is watching
TVo The first step is that some speaker S verbalizes this in an
utterance to some listener{s) L. For our present purposes we
may call this non-reported speech. Reported speech occurs when
a speaker verbalizes the situation given in section 1 and
renders it eiLher in the form of direct re ort (section 2a) or
in the form of indirect report {section 2b I
y
O.
•
INITIAL SITUATION
[Eo is watching TV]
1
NON-REPORTED SPEECH
[S utters to L]
•
Eo is watching TV
DIRECT REPORT
2a
[S utters to L]
Tom says
"Eo is watching TV"
INDIRECT REPORT
2b.
[S utters to L]
Tom says
that
0
Eo is watching TV
Table 1. A first illustration of speech reporting. Speech
situations are 'indicated in square brackets. S
a speaker,
L
a listener.
=
=
The little model in table 1 can be expanded and varied
in several waysr of course. It may be applied recursively:
John says that Tom says that Eo is watching TV, etc. Statement,
question and command may be combined in various ways. So may
the identity of the persons involved. Time and place relations
may vary. Etc.
One example of a particular combination is given in
table 2. What interests us here is the difference in form
between direct and indirect report. We notice that the indi­
rect form in this example brings about a change' in word order,
the insertion of a subordinating particle if, and the altera­
tion of person, tense , and punctuation.
-2-
1.
NON-REPORTED SPEECH
[ Tom utters to Fred] Was Bo watching TV?
2a.
DIRECT REPORT
[ Sven utters to Bo ]
Tom asked Fred
2b.
INDIRECT REPORT
Tom asked Fred
[ Sven utters to Bo ]
,"wa'
R
tChing TV? "
.if you had been
watching TV.
Tabel 2. Example based on table 1.
The reason for these changes is obvious: in the direct
report, the lower clause still retains its deictic reference
to the situation in which it was originally uttered, the situa­
tion in section 1 in the table. In the indirect report, the
lower clause is governed deictically by the higher clause and
is viewed in relation to the situation in which the higher
clause is uttered ( io e. the situation in section 2b ) . Thus, Bo
is third person in Tom 's utterance to Fred, but second person
when he is addressed by Sven. Tom 's utterance is a direct
question to the listener Fred, but not to the listener Bo.
The forming of indirect report thus involves a sentence
subordination process and a number of other adjustments deter­
mined by the change in deictic reference. What we want to find
out here is ( a ) to what extent various languages actually
distinguish between a direct and an indirect form of report,
and ( b ) what grammatical means they apply. We shall be con­
cerned with Swedish in comparison with the ten immigrant
languages that are involved as so urce languages in the SSM
error investigation. 2
Indirect report is a nice field to demonstrate the work­
ings of subordination in languages, and it is also worthy of
attention because it is an area of Swedish grammar where
learners run into many problems. Report is, however, not a
syntactic clas s but ra.ther a functionally defined sentence­
type. There is not a clear borderline between indirect report
and other types of hypotactic sentences. Most typically, we
have to do with a quotation clause that is governed by a verb
of saying (�, tell, ask, report, write etc. ) . This is the
type of sentences that we have investigated in our sample of
immigrant languages. Further, the verb may be nominalized ( the
guestion whethero . o ) , or impliedo We shall not deal here with
cases where the quotation comprises longer stretches of sen­
tences, nor with the additional markers of report that may
occur, such as according to our spokesman and the like. But it
is worthwhile to include examples of reported thought, i. e. ,
quotations governed by a verb of thinking ( think, believe,
assume, know, remember etco ) since these sentences are con­
structed�ctly like reported actual speech in Swedish and
present the same learning problemso In languages like Swedish
a criterion to show that a subordinate clause can be understood
indirect report is that it is possible ( with reasonably ac­
as
ceptable result ) to supply a "direct" counterpart For example,
one may try to judge in sentences like (1) and ( 2 ) if a verb
0
-3-
can be used as a verb of report.
sa
tankte
Lasse ? antog : "Peter ar tokig".
? visste
l
Lasse
I
I
fragade
undrade
Peter
:?ifragasatte : "Ar
tokig
?"
? visste inte
I
Lasse said I thought/
assumed I knew, "Peter
is crazy".
L asked I wondered I
called in question I
didn 't know, "Is P cr?"
A group of sentences that are abundant in the error
material consists of what we may label "pseudo-reports". These
are sentences like I 'll tell you what happened, He explained
how the trap worked, I remembered what Ann had said. They look
like indirect reports in that a clause with the form of an
indirect wh-question is governed by a verb of saying or think­
ing. Yet the lower clause is not a quotation, and there is no
"direct" counterparto With an inquiring verb they could pass as
indirect reports: I ask you what happened, He asked how the
trap worked, I wondered what Ann had said. Pseudo-reports cause
the same types of learning problems in Swedish as true indirect
reports, judging from the error material, and they involve
largely the same grammatical mechanismso Hence, we shall give
them some attention in the followingo
If the direct and the indirect report form were always
equally likely to occur, we would always have to rely on overt
markers in order to tell them apart. But this is not actually
the case. There are many contexts, styles, and types of text or
discourse where the indirect form is clearly the dominant, or
most expected one. This would apply, for instance, to most
contexts in a reasoning text. Since we are primarily interested
in the system of distinctive markers here, we shall mostly
leave the bias of the context aside. But still, there are cases
where the interpretation of direct or indirect has to be deter­
mined from the context in order, for instance, to identify a
deictic referent, as in the Finnish sentence (3)0
Imatti kysyy tuletko
sinal Matti asks,
"Are you coming?"
Matti aska come-QUESTION you
or
. o . if you are coming.
The person is adjusted in the indirect form in Finnish. Unless
Matti and the reporting speaker address the same listener,
ISina/ 'you' will have different reference depending on whether
the sentence is understood as direct or indirect report. The
interpretation has to be derived from the context.
Now let us look at the various mechanisms that are used
to form indirect report. Error data will be used to expose
these mechanisms in Standard Written Swedish an4 to demonstrate
that they constitute problems for foreign learners.
Subordination mechanisms
Various means of subordination are used to form indirect
report in the languages that we have looked at. Most of these
grammatical means are employed in one way or another in Swedish,
Ao
-4so one way of putting together examples of them is to give a
brief survey of direct and indirect quotation clause structure
in Swedish. This is done in table 3 and sentences (4) to (11).
INDIRECT
DIRECT
V
Stat ement
#
Command
# V.lmp
X
(4)
( 5)
# att S V
# att S V
att V.lnf
Yes-No Question
# V S
(6)
# om S V
Wh-Question
Wh � S
Wh = S
# Wh V
# Wh V
(8)
(10)
# Wh S V
# Wh som V
(9)
(11)
Table 30 Structure of direct and indirect quotation clauses in
Swedisho Explanations:
# = clause boundary
= an unspecified constituent
X
S
sub ject
V = finite verb
V,:l.mp= verb in the imperative
V.lnf= verb in the infinitive
Wh = interrogative word
( ) figures in parentheses refer to examples in the text
=
IIRolf ar hemma idago ll
IIIdag ar Rolf hemmao "
(5) Lisa sager, att Rolf ar hemma
idago
(6) "Ar Rolf hemma idag?"
(7) Lisa fragar, om Rolf ar hemma
idago
(8) "Varfor ar Rolf hemma?1I
IIVem ar Rolf?1I
"Vad sager Rolf?"
(9) Lisa fragar,
varfor Rolf ar hemmao
vern Rolf ar.
vad Rolf sagero
(10) "Vern ar Rolf? II
IIVad hander hemma?"
(11) Lisa fragar,
vern som ar Rolfo
vad som hander hemmao
(4)
"Rolf is at home today.1I
Lisa says that
• • •
IIIs Rolf at home today?"
Lisa asks if. "o
"Why is Rolf at home?"
"Who is Rolf?"
IIWhat does Rolf say? 1I
a
a
awho Rolf
is.
"Who is Rolf?"
IIWhat is happening at home?"
a
•
•
which
one is
Rolf.
-5Word order
The sentences in (8) and (9) illustrate the role of word
order in forming such indirect wh-questions in which the wh­
word is not the subject. They are parallelled in the error
material by examples like (12) and (13) in which the learners
have used the word order of the direct question. ( Here and in
the subsequent error examples, a represents the learner 's
version and b a Standard Swedish version with an English word­
by-word translationo The native language of the learner is
given in brackets. Underlinings are ourso )
Alo
ao *Jag har en granne som ringer varje morron och fragar,
hur mar jago [ Finnish ]
som ringer varje morgon och
bo Jag har en granne
I
have a neighbor who calls every morning and
fragar hur ja� mar.
how I
feel.
asks
0
(13) a. i�fragar de inte efter vad heter �ranneno [ Japanese ]
heter.
bo fragar de inte efter vad grannen
what the neighbor is calledo
they don 't ask
The same solution occurs in the case of pseudo-reports,
as in (14).
(14) a. *Han forklarade hur uppvarms husen med V::l.rmt vatten.
[ Greek ]
med varmt
uJ2J2varms
Han forklarade hur husen
He explained how the houses are heated with hot
vatteno
water.
Another marker of subordination is the placement of
sentence adverbials ( including the negation ) before the finite
verb, as opposed to the placement after the fjnite verb in main
clauseso If we negate the examples (4) - (11), we get "Rolf ar
inte hemma idag" versus att Rolf inte ar hernma idag, "Ar Rolf
inte hemma ida ? " ("Ar inte Rolf hemma ida ?") versus om Rolf
inte ar
om inte Rolf ar 000
and so ono The, place of
sentence adverbials is a clear crlterion of subordination in
the case of questions ( examples (6) - (11))0 For statements,
the distinction between direct and indirect on this basis is
less clear-cut since certain types of att-clauses may optional­
ly ( and do in fact frequently ) occur with main clause word
order especially in colloquial style ( cf o Teleman 1967, Lind­
berg 1973, Andersson 1975). This gives us sentence variants
such as Lisa sa er att Rolf ar inte hemma as an alternative
negative counterpart to sentence 5 . Likewise, Swedes will
hardly consider a sentence like (15a) as an error.
0 0 .
,
(15) ao Jag tror att jag kan inte glomma det rummeto [ Spanish ]
bo Jag tror att jag inte kan glomma det rurnm�to
not can forget that roomo
think that I
I
-6In more formal writing, there is a stronger inclination to
reject ( 15a ) and require ( 15b ) 0 Another natural way of ex­
pressing this sentence is with negation raising: Jag tror inte
( att ) jag kan glomma det rummet 'I don't think ( that) I can
forget that room'o
It is also possible to find cases in the error material
where the Verb-Negation order appears very awkward, as in
example ( 16 ) 0
( 16 ) ao *Hon tanker att jag vet inte vad hon gor. [ English ]
bo Hon tror
att jag inte vet vad hon gore
She thinks that I
not know what she is doing.
The corresponding direct report would be Hon tror: "Mamma vet
inte vad jag gor"o Notice that sentence ( 16 ) involves person
adjustments characteristic of the indirect form, Mamma- � md
�
hono This may explain why ( 16a ) is harder to accept than
( 15a ) 0 What is essential about att-clauses for our present
problem is that even if some of them allow main clause word
order, the word order of ( 15b ) is still available and natural,
even in colloquial style, to help distinguish indirect report
from direct. Yet, it is obvious that att-clauses form a point
in the Swedish system where this distinction, so clear-cut
elsewhere, tends to get neutralized. Teleman ( 1967,167 ) quotes
sentences from his spoken corpus in which att introduces
quotations where the adjustment of person and tense has not
taken place; as he remarks, --att here functions like the colon
in writingo
-
Place holders
In an earlier paper ( Hammarberg & Viberg 1975 ) we have
discussed the role of syntactic place holders in Swedish and
related languages. One type of place holder that plays a part
in connection with reported wh-questions in Swedish is the
subject particle som that is inserted in cases where a wh-word
(�, vilken,� )� the subject of a subordinate clause, ioe.,
precisely in those cases where the word order ( Wh V ) would
otherwise signal a direct question ( cf. Beckman 1922 ) 0 Examples
of this are ( 10 ) , ( 11 ) , ( 17 ) and ( 18 )
'
( 17 ) a 0 * och fragade vilken film var dar [ Spanish ]
bo och fragade vilken film som gick
dar
and asked
what
film
was on there
A2.
0
( 18 ) a. *Vi maste tanka pa vilka behallander passar bra for
demo [ Japanese ]
0
b. Vi maste tanka pa vilka forhallanden som- passar bra
We must think of what conditions
are suitable
for demo
for them.
0
Subordinating particles
A subordination marker that normally occurs in sentences
like ( 5 ) and ( 7 ) is the subordinating conjunction, att 'that'
AJo
-7with statements and om 'if',
sider examples ( 19 ) �d ( 20 )
learner's version.
'whether' with questions. Con­
where att is missing in the
��tyvarr maste man saga det finns inget perfekta
jamlikhet [ Japanese ]
maste man saga att
det
finns ingen
tyvarr
unf�ately must
one say
that there is
no
perfekt jamlikhet
perfect equality
tyvarr maste man saga att det inte finns nEt.gon po
any
not
( 20 )
j.
a.
��Jag hoppades, det har gangen kunde jag lasao
[ Japanese]
b.
Jag hoppades, att
den har gangen skulle jag kunna
I
hoped
that this
time
would
I
be able to
lasa.
read.
Jag hoppades
( att )
jag skulle kunna lasa den har gangen.
Certainly � is not always obligatory, but can be left
out in some types of sentences. We cannot digress into the
problem of att-Iess clauses here, but some relevant discussion
is found in Teleman ( 1967,169 ) , Thorell ( 1967,236 ) and
Andersson ( 1975,212ff. & 229 ) .
By contrast, the error material shows various ways of
using � in the wrong context. There is a tendency towards
overusing � as a marker of embedded quotations, and also to
use it with pseudoreports.
It is of course only with finite clauses that a distinc­
tion of direct/ indirect can be expressed; but considering what
was said above about att functioning like a colon, it is
natural that we should find examples such as ( 21 ) .
aoo i�Hon prukar saga ofta att "jassa", men jag tror hon ar
inte forstod prisis va�ag menaro [ Finnish ]
bo Hon brukar ofta saga "jassa",
men jag tror
att
hon
She often says
"indeedl" but I
think that she
inte har forstatt
precis
vad
jag menaro
not
has understood exactly what I
mean.
Att used as a question embedder is shown with a yes- no
question-in ( 22 ) and with a wh- question in ( 23 ) 0 The structure
of the learners' sentences is att + Direct question.
( 22 )
ao
��En fraga till att bar aktenskapet avskaffaso
[ Hungarian ]
b.
En fraga till
ar om aktenskapet bor
avskaffaso
Another question is if marriage
should be abolished.
-8i"vi kan inte sager att vad ar skilnad melan Sverige
och Irano [ Persian ]
bo Vi kan inte saga vad skillnaden
ar mellan
We can not tell what the difference is between
Sverige och Iran.
Sweden and Irano
Vi kan inte saga vad som ar skillnaden
---
.
0
0
In (24) att is combined with a wh-question showing the
S V order of a subordinate clauseo
(24) ao i�an vet inte att var gransen star. [ Turkish ]
b. Man vet
inte var
gransen
gar.
One knows not where the borderline goes.
(25) shows an alternative solution to (22) : a yes-no
question is embedded without the �, but retains the V S order
of a direct question.
-!�ga titta den nya vaningen att jag vet passar mina
gamla gardiner dit. [ Finnish ]
ga och titta pa den nya vaningen sa att jag vet
go and look at the new apartment so that I
know
om mina gamla gardiner passar dit.
if my
old
curtains fit in there.
Nominalization
In some languages, indirect report is among those sub­
ordinations that are regularly done by means of nominalization.
The type of nominalization that we have in mind here results
in constructions like 'She told me his eating' for 'She told
me that he was eating ', or 'Lisa said Rolf's having been at
home ' for 'Lisa said that Rolf' had been at home '. The verb of
the lower clause is turned into a participle with the subject
of the lower clause as a genitive modifier. In our language
sample, this applies in Turkish where it is the usual way of
subordinating sentences, and in Finnish where it occurs with
indirect statements as an alternative to a finite, that-clauseo
Nominalization is not a device for indirect report formation
in Swedish.
We have found no clear evidence that this kind of nomi­
nalization may influence learners of Swedish. The only possible
instance in our error material so far is the Turkish learner 's
use of the genitive kungens in (26 )
A4.
0
(26) a. irDu fragar om vad kungen� hette. [ Turkish ]
b. Du fragar
(om)
vad kungen
hette.
You are asking (about) what the king was called.
A50
Intonation, punctuation etc.
Intonation in speech and punctuation, capitalization etc.
in writing serve as phonic and graphic markers of direct or
indirect report. In an error example like (27) , theIDrd order
-9-
and the marks of question and exclamation are signals of
direct report whereas the person adjustment, the small � of
vad, and the absence of a colon and quotation marks express
indirect reporto
a
��De star och tanker vad ska de gora att oppna kassan!?
[ Persian ]
bo De
star och tanker pa
vad de
ska
gora
They stand there thinking about what they shall do
for att
oppna kassano
in order to open the cashier's officeo
The corresponding direct report with the graphic markers
meticulously included, would run De star och tanker: IIVad ska
vi gOla. for att oppna kassan!?"
0
Bo
Shift of mood
The use of the subjunctive mood to express "indirectness"
has its well-known and interesting manifestations in German, in
cases where "indirectll implies tlnot necessarily factual from
the point of view of the present speaker"o The use of mood to
express "indirectness" deserves to be noticed here as a
structural device available to languages Its role in Swedish
is very limited, however The subjunctive of � 'be' and
strong verbs is still possible in officialese writing in con­
texts like Ordforanden menade att fragan vore tillrackligt ut­
� 'The chairman was of the opinion that the issue was
sufficiently penetrated'o The use of the modal auxiliary skall,
skulle occurs notably in reports without a governing verb,
eo g., in a rendering of a witness report like Mannen skulle ha
fardats i en skar Toyota 'The man was said to have travelled
in a pink Toyota'o
Another shift of mood is the elimination of the impera­
tive in embedded commands, as in Jag sa at honom att han skulle
vara forsiktig II told him that he should be careful' versus
Jag sa at honom: "Var forsiktig�"
0
0
Co
Deictic adjustments
In a way, deictic adjustments such as shift of tense,
person, or place reference, are only secondary markers of
indirectness As we tried to show with example (3), a basis
for the interpretation of the deictic reference has to be
given by the context and/or by the morpho-syntactic means that
we have discussed in sections A and Bo But provided this, the
deictic markers strengthen the contrast between the direct and
the indirect form, and the extent to which a language uses
deictic adjustments tells us something about the importance of
the direct/indirect dichotomy in that language
0
0
Time
Swedish is among the Languages in which the sequence of
tenses distinguishes indirect report from directo Errors in
this respect are common with learners from several language
areaso Here are some examples
C 10
0
-10-
(28) ao * Gubben fick veta att han ar valkommen. [ Polish ]
is
fick veta att han var valkommen.
bo Gubben
The old man was told that he was welcome.
(29) a. * I den har tid medelade tlNobelfredscomitetl att de
[ Arabic ]
beslutade att
decided
bo Vid den har tiden meddelade Nobelpriskommitten
time announced the Nobel Prize Committee
By this
att den hade beslutat att
that it had decided to
o
.
0
. 0 0
(30) a. *Han sa att de ska lyfta henne sa att hon kan krypa
will
can
in i lockano [ Persian ]
bo Han sa att de
skulle lyfta henne sa att hon kunde
He said that they would lift her
so that she could
luckano
krypa in genom
creep in through the hole.
0
(31 ) ao *Jag tankte att, jag ska
WIT 1
ga till Kostgallerio [ Finnish ]
0
b. Jag tankte, att jag I;lkglle ga till Konstgalleriet.
I
the Art Gallery.
thought that I would go to
0
There are various misinterpretations that may arise in
Swedish from the failure to adjust tense, unless the context
secures the right interpretation. One effect may be the im­
pression tllat the reporting speaker explicitly acknowledges
the contents of the quotation as something 'factual' and
'still valid', as in Lisa told me that Rolf is spelled with a
capital R; cf. (28) 0 Or the time sequence between the higher
and the lower predicate may be effected� cf. Lisa told me that
she loved Rold versus had loved Rolf; cfo also (29). Still
another effect is merely that the indication of direct/indirect
report is blurred.
(32) and (33) give examples of shift in time adverbialso
(32) Lisa berattade for mig: "Rolf ar s j uk idagtlo
, att Rolf var s j uk den dageno
Lisa told me, "Rolf is sick today"
that Rolf was sick that day.
0
(33) SITUATION:
[X tells Y on Wednesday [ Lisa tells X on
Tuesday [ Rolf will come on Wednesday ] ] ]
DIRECT:
Lisa berattade for mig: "Rolf kommer i mo�gon"o
tomorrow
INDIRECT:
Lisa berattade for mig att Roll' skulle komma (�) idago
today
-11-
�ome Swedes use an optional som 'like' colloquially in ex­
pressions like som idag, som-pI fredag 'on Friday'0 It is
clear that the function of this � is to point out that the
time expression has been deictically adjustedo
Place
Place reference may be shifted by means of place ad­
verbials, as in (34)0
C20
Lisa sa i telefon
"Vi har en fest och Rolf ar ocksa har"o
att de hade en fest och att Rolf ocksa var dar.
Lisa said on the phone
"We are having a party and Rolf is here too"o
that they were 000 and that R was there tooo
Error examples are rare; here is one:
(35) ao *Men han glHmde inte fHr vad da kom vi hito [Polish]
here
fHro
bo Men han glHmde inte vad vi hade kommit dit
there foro
But he forgot not what we had got
Person
(36) and (37) are error examples where the adjustment of
person is missingo
C30
(36) ao *Hon fragar dem att vad ska vi gHrao [Japanese]
bo Hon fragar dem vad de
ska
gHrao
She asks
them what they shall doo
*En dag sa Ante till sin fru Sonja att � inte
trivdes bra har som i Laxentrasko [Persian]
bo En dag sa
Ante till sin fru Sonja att han inte
One day said Ante to
his wife Sonja that he not
trivdes lika bra har som i Laxentrasko
liked it as
well here as in Laxentrasko
Cross-language discussion
After this look at the various markers of indirectness
and the Swedish side of the problem, let us include the other
languages and try to get a comparative viewo The discussion
will have to concentrate on a few selected points that will
show alternative structural solutions in immigrant languag�so
Data from the languages will then be summed up in table 40j
Word order
In Swedish word order plays a crucial role in signaling
the distinction between main and subordinate clauses and thus
between direct and indirect speecho This function of word
order, together with its function to mark the difference
between a direct statement and a direct yes-no-question, is
found only in Swedish and a small group of related languages
(the Germanic languages ,Hid to some extent French) 0 We have
- 1 2-
discussed this in Hammarberg and Viberg (1975). Indirect wh­
que. tions were briefly discussed in that report. But word
order in indirect wh-questions is a great problem to most
immigrants learning Swedish, and the background of errors dis­
cussed in sections A1 and A3 above is rather different in
different immigrant languages. Therefore we shall present a
more detailed typology for word order in indirect wh�questions
to complete the picture given in the earlier report. It is con­
venient to make the charaoterization in two steps:
(a) The extent to which the contrast between direct and
'indirect wh-questions is signaled by word order.
(b) How the contrast is made by word order.
In Swedish the basic mechanism for making the contra�
involves two parts:
Wh-words are initial.
The position of the subject is strictly grammati­
calized and differs in direct and indirect questions
(VS/SV), a special case being when the subject is
questioned.
It might be sensible to ask whether there is any
other way of making the contrast than by inversion,
and whether inversion always implies that there is
a contrast.
Among the languages under discussion in this paper, only Engliffi
belongs to the same type as Swedish:
Wh V S
(3 8)
'
What is Peter doing?
(39) I wonder what Peter is doing.
Wh S V
The use of the so called operator do in direct questions is
also important in marking the contrast between direct and
indirect questions by word order:
Wh V S
(4 0) What did Peter buy?
Wh S V
(41) I wonder what Peter bought.
English is not as consistent as Swedish, though. There is no
contrast in word order when the subject is questioned:
Swedish (cf. table 3)
(42) Who is coming?
Wh V Wh V
(43) I wonder who is coming.
Wh V Wh som V
Due to the place holder �, Swedish maintains the contrast
even when the subject is questioned (see examples 1 0, 11 and
17, 18 above) .
There is no contrast in word order between direct and
indirect wh-questions in any of the other nine languages, at
least not a consistent one. This is one possible reason why
word order in indirect wh-questions is a prob lem to most
learners. They are not used to making a contrast by word order
in their own language. Viewed more in detail the background is
rather different in different languages. In order to demon­
strate this, the languages in our sample can be grouped into
the following types:
I
Languages that do not obligatorily put wh-words in
initial position (several subtypes) .
Japanese, Turkish, Persian, Hungarian
-13-
II
Languages that obligatorily put wh-words in initial
positiono
This type may be subdivided as to whether there is or is
not an inversion of subject and verb after an initial wh­
wordo
ao Languages without an obligatory inversiono
In this subtype the position of the subject is not
grammaticalizedo In the unmarked case the word order
is SY but the subject may often occur somewhere after
the verb, too. The choice of position seems to be
governed mainly by textual factors. There is no con­
trast in word order between direct and indirect wh­
questions, at least not a consistent oneo
Fblish, Finnish
bo Languages that obligatorily put the subject after the
verb in both direct and indirect questionso
The position of the subject is not completely bound
in this subtype, eithero But it must not occur in
front of the verb in a wh-question ( provided it is
not the subject that is questioned ) . Most often the
subject occurs directly after the verb or verbs"
Spanish, Greek
co Languages with an obligatory inversion in direct
questions but without any inversion in indirect
questionso
Swedish, English
The languages of type I are all S OY except Hungarian
which is today basically SYO with some reminiscent traits of
SOYo The languages of type II are all SYOo ( Cfo Greenberg 1 963,
Universals 1 1 , 12 0) Only type II c uses word order to make a
distinction between direct and indirect questionso In type II a
an inversion is not obligatory in either case and in II b in­
version is obligatoryo This difference between I, II a and II b
is part of the background when it comes to explaining learners'
errors like (1 2), (1 3) and (1 4) 0
Particles and nominalization
The subordinated status of a clause is often indicated
by morphological or lexical markerso In our sample of languages,
three different mechanisms may be distinguished: The use of
clause initial particles, the use of clause final particles,
and the use of nominalizationo
An important characteristic of a subordinate clause in a
language of the first type is that it is introduced by a
particle that belongs to a rather restricted set of grammatical
words Swedish and most of the other languages in our sample
belong to this type. Many subtypes may be established that�e
relevant to the discussion of the errors in section A3 above.
Persian, for instance, is characterized by using the same
inicial particle ke 'that' to embed both statements and
questions Compared to Persian, Swedish has a rather rich
system of different clause introducing particleso Another
relevant factor is whether a certain particle tha.t introduces
0
0
-14-
a subordinate clause may introduce a main clause as well.
the latter is the case,
then the particle by itsel� is no
clear mark o� subordination. In Swedish,
'i�'
I�
the particle �
e.g.,
in an indirect question is a clear mark o� subordination,
but a wh-word is not.
The second major type is represented by Japanese:
(44)
Jiro
wa
Taro ga
doa
aketa
0
to
iu
Jiro THEME Taro SUBJ door OBJ opened QUOTE says
Jiro says that
Taro opened the door
The quotation particle to may be used in all kinds o� reported
speech, both direct and indirect, questions as well as state­
ments. No clause introducers are �ound in Japanese.
(1974)
has noted,
this is in accordance with the
question words are not normally placed in initial
that relative pronouns are completely lacking.
sistently marks subordination clause-�inally.
what is to
be expected in
The third
an SOY-language.
major type is
As Kuno
�act
that
position and
Japanese con­
This is o� course
represented by Turkish.
In this
L.nguage all types o� indirect speech may be expressed by a
nominalized clause. The �ollowing examples give an
idea o� how
an indirect statement is �ormed:
(45)
Ali
(46)
Ahmet Ali-nin
gel
-di
Ali came
come-PAST
-GEN
geldi�ini soyledi
'coming'
said
Ahmet said that Ali had
come
The same kind o� nominalization is used in both
direct questions.
according to a somewhat di��erent
su��ix on the
types of' in­
Indirect .commands are nominalized,
too,
pattern and with another
verb. In addition to this pattern,
but
another pattern
exists
in Turkish with an inveriable initial particle, ki.
.
Most languages belonging to the �irst type seem to-be
�airly consistent in having
o� indirect speech.
Swedish,
English,
some initial
particle in
all types
This seems to hold true at least �or
Spanish, Greek, Polish and Persian. Japanese
is very consistent,
too,
in always using clause �inal particles.
But o� course there may be languages that are not consistent
and where the di��erent types of' indirect speech �ollow
dif�erent patterns. In Turkish there are two di��erent possi­
bilities, at least in some cases. But the language is consistent
in allowing some kind o� nominalization in all· kinds o� in­
direct speech. In our sample, Finnish is to a rather great ex­
tent a mixed type. In indirect statements two di��erent con­
structions may be used. Either an initial subordinating par­
ticle etta 'that' is used, or, as a second alternative, the
clause is nominalized:
(47)
Olavi sanoo,
says
etta
that
Pekka kuorsaa
snores
Olavi says that Pekka snores
(48)
Olavi sanoo
Pekan
-GEN
kuorsaavan
'snoring'
Olavi says that Pekka snores
-15-
But in indirect yes-no-questions nominalization is not used.
Usually such questions do not have an initial particle either,
but have the same form as a direct question:
Does Pekka snore?
(49) Kuorsaako Pekka?
snores-Q
Olavi kysyy, kuorsaako Pekka. ( Q
QUESTION MARKER )
snores-Q
asks
Olavi is asking if Pekka snores.
In what is usually considered as substandard, yet very common
speech, indirect questions may be introduced by the particle
etta 'that' in front of the verb with the question particle.
Another possibility is to use a conditional �lause introduced
by jos 'if'. Then the indirect question is constructed exactly
as in Swedish. Actually, this last type is characterized by
Finns as a Swedish trait and was rather often totally rejected
by our informants.
(50 )
=
Deictic adjustments
In Swedish as in all the well known languages of Europe
such as English, German and French, indirect speech is accom­
panied by a series of deictic adjustment of time ( adverbials,
tense ) , person and place ( here /there ) . In many of the immigrant
languages some of these adjustments do not occur or at least do
not occur obligatorily. In order to illustrate the type of
languages that lack some of the deictic adjustments, we have
chosen to present data from the two languages in our sample
that are most dissimilar to Swedish in this respect. For ease
of reference the phenomena are discussed language by language.
At the end of this section there is a short summary mechanism
by mechanism that mirrors the presentation in the first part
of our paper ( table 4) .
The following data were collected from a Persian infor­
mant. What is given in square brackets represents the intended
speech situation that was first described verbally to the in­
formant.
Non-reported speech
[ Ali to Hassan: ]
(51 ) ( aja ) ( tu ) khasteh hasti
Are you tired?
Q
you tired
are [2 pers sing, present ]
Direct report
[ Hassan to Moh ammed: ]
(52) Ali sual
mikonad
( aja ) ( tu ) khasteh hasti
are
question is-making
you tlred
Q
Ali is asking, "Are you tired?"
Indirect report
ke
mikonad
( man ) khasteh hastam
Ali sual
question is-making that
all!
I
tired
Ali is asking if I'm tired.
-16We also asked the informant what the indirect question would
be if reported at a later time than in a direct connection to
the original speech situation like:
kard ke
(man) khasteh hastam
(5 4 ) Ali sua1
I
am
tired
question made that
In all the different combinations of persons and tenses that
we tried on the informant the result was like this one . The
informant adjusted the person but preferred not to adjust the
tense. In all cases, however, he gave a form with adjusted
tense as a second alternative, or at least he recognized this
as a possible alternative when explicitly asked about that . As
a second alternative to (54) above he thus gave:
(5 5) Ali sual
kard ke
(man) khasteh bodam
tired
was [1 pers sing,
question made that
I
past]
According to Lambton (1974, 15 5 ) even person may in some cases
be left unadjusted although we were not able to obtain such
forms from our informant:
'Aj"ter verbs of saying, thinking, knowing, seeing, etc. , the
tense of the verb is normally that of direct speech, but the
pronoun is not necessarily that of direct speech. '
Japanese is another language that uses deictic adjust­
ments to a lesser degree than Swedish .
Non-report
Is my face dirty?
(56) boku kao ga yogorete iru kai
[male speaker ]
I
face SUBJ dirty
is Q
Direct report
(57) Taro wa
boku kao ga yogorete iru kai to
kiita
Taro THEME I
face SUBJ dirty
is Q
QUOTE asked
Taro asked, 'Is my face dirty?'
Indirect report
jibun no kao ga yogorete iru ka (to) kiita
(58) Taro wa
Taro THEME self 's face SUBJ dirty
is Q QUOTE asked
Taro asked if his face was dirty
As can be seen from (58), tense is not adjusted in Japanese .
This holds true in general . But person is not always left un­
adjusted. In (58), reflexivization works as a kind of person
adjustment . The fact that personal pronouns often are left out
and that the verb is not inflected for person makes person
adjustment less prominent . The levels of politeness, a very
characteristic trait of Japanese, are neutralized in sub­
ordinate clauses . This might be of some importance for signal­
ing the distinction between direct and indirect report . But in
the main, adjustments are not as common in Japanese as in
Swedish and the distinction between direct and indirect report
is often not very sharp in Japanese .
0
20
3.
1
etc
)
4.
of
a
+
+
++
+
+
+
+
+
th at
+
(+)
+
+
tf}
�
t:lt
::r:
tf}
H
reporto
+
+
+
+
ffi
�
ril
ril
the
+
+
t:lt
�
�
H
tf}
+
+
+
�
�
Q
H
+
+
+
+
+
+
H
�
�
::r:
tf}
H
c ase to
a
the
�
+
+
+
�
�
E5
::r:
tf}
+
( +)
+
(+)
�
�
t:lt
�
ril
tf}
ril
significant degree.
some
of categories
+
+
+
+
$
0
z
�
H
langu age m ake s
classific ation
that
The
+ �+))
+ +
+
+
+
::r:
tf}
H
...:l
0
t:lt
plus indicates
this is not
A
+
++
+
+
+
+
::r:
tf}
H
...:l
0
Z
ril
indirect
texto
forming
the
minus
sections in
mech anisms
to the
of the mechanism,
refers
Summary
Person
of
adverbials
Place
tense
ao
adjustments
b.
Time
Deictic
of indirectness
mood:
subjunctive
of
punctu ation
bo
Shift
intonation
graphic markers
p articles
ao
and
Phonic
5.
final
Nominalization
initial
aD
bo
Subordin ating
pronouns
subject
subject particle
ao
bo
regul ar use
A2
mech anisms
holders
order
Pl ace
Word
4.
3.
10
20
Subordin ation
Table
c.
B.
Ao
�
tf}
::r:
tf}
H
�
(A1 ,
-...]
I
-I
-18-
SAMMANFATTNING
Anforing
i svenskan
Att
en
och
Uppsatsen undersoker
invandrarspraken.
I
1
och 2
( icke-anforing ) ( avd 1),
anforing.
Vad vi
och
f�lle
styrs
( t�nka,
hand pa
anta,
veta etc
�foringar
Vi
ska s�ga
att en
fran
introducerar termen
dig
sats
vad
sadana
som h�nde.
)
de
bisatsen
ett
/ ind irekt
(A)
�r
justeringar av
att svenskan
tids-,
�r
ett
"indirekthetsmedel".
D�refter tas de
och
som
3
och
olika
kan ha
exemplen
och
starkt
subjekt
av
iska
och japanska.
tabell
4.
bakgrund
framfor-
nominalisering
en
fran svenskan
att
(C)
roll
bilda
in­
som ordfoljd,
intona­
deiktiska
I
stort
g�ller
mycket av
dessa
i
bilden
for vilken
och
verb
)
for
roll
spelar
och indirekt fraga
hur
sinsemellan
sprakomraden
i de olika
lik­
( jfr
ex
spraken.
och efterst�llda underordnings­
j�mfors.
j�mforelse med
genom att gora
Data
med
En typologi ges
kan belysa t ex
varierande
exemplifieras genom
styrs av
(4) - (11) .
invandrarspraken
av frageord,
Detta
Sprakens anv�ndning
partiklar och av
Jag
s�tt
Kontextens
extremt
artade sprakfel hos invandrare fran olika
13, 14)
be­
som
pa sa
form
personsyftning.
anv�nder
oversikt.
( placeringen
spraken.
fall
SSM:s felmaterial
for att signalera skillnaden mellan direkt
12,
for
),
t�nke­
Strukturen hos anforda satser i svenskan
sammanfattas i tabell
att ge en komparativ
etc
Problemen
nominalisering,
modusv�xling
plats-
sprak
skriva
d�r ett
inte ett citat eller
citato
svenskan och
(B)
rapporterad
citatsats.
satsunderordningsmekanismer
interpunktion,
till­
anforingsform diskuteras.
underordningspartiklar,
platshallare,
i de elva
fall
olika grammatiska mekanismerna for
direkt anforing:
ordfoljden
en
ett
invandrarspraken.
De liknar anforingar
Med exemplifiering ur
tion och
tanke",
vidare
som
fraga,
pseudo-anforing
men
identifiera direkt
genomgas
och
fall d�r en
ber�tta,
styr
det
andra hypotaktiska meningar
ens en sammanfattande parafras av
for att
och in­
direkt
tar detta
med den indirekta frageordsfragans
ett s�ge- eller t�nkeverb,
indirekt
och
en del av
vi
i svenskan
"rapporterad
att s�rskilja
rors.
och
( s�ga,
ett s�geverb
tro,
icke-rapporterat
och
som da anv�nds.
indirekt anforing
i forsta
av
)
i
mellan
i vilken utstr�ckning
dels
medel
underordna satser,
men vi inkluderar �ven
verb
vilka
att belysa bisatsbildning
Vi ser
sats
2a-b
i uttrycksformen mellan
dels
Syntaktiskt �r
problemet att
pa �r
varierar
visas skillnader
( avd
�r
sprakbrukare.
samt mellan direkt
svar
saker
spraken gar en skillnad
direkt anforing,
som
om uttrycksmedlen for detta
tabell
eller anforing
rapporterat tal
tal
infor
st�lls
alIa
vi
sagt,
aterge vad nagon
t ex att
tal,
rapportera
vanlig uppgift som
invandrarsprak
tio
for de elva
fa
tva
Deiktiska
sprak
sadana,
som
justeringar
avviker
n�mligen
pers­
spraken sammanfattas i
-19NOTES
*
1
2
3
This report is an expanded version of a paper given at
the Third Scandinavian Conference of Linguistics, Hana­
saari, Finland, October 1-3, 1976. It also appears in
the volume of Papers from that conference, ed. by Fred
Karlsson, published by The Text Linguistics Research
Group, Academy of Finland, Turku 1976.
It would seem natural to refer to non-reported speech as
'direct speech', but this term has traditionally been
used for our concept direct report in contexts where no
explicit distinction was made between our case 1 and
case 2a. See e.g. Jespersen 1924, 290.
Written errors by Arabic, English, Finnish, Greek, Hungar­
ian, Japanese, Persian, Polish, Spanish, and Turkish
learners of Swedish are taken into account in the error
investigation which is a part of the SSM project. See
Hammarberg (1975) for a general presentation. An extensive
account for the error study is in preparation.
We base this section in part on work with native infor­
mants, in addition to standard grammars. And not least,
we owe thanks to many colleagues around us whom we have
approached now and then to get points in "their"
languages clarified. Table 4 is not quite complete yet;
we do not necessarily intend to leave it at that.
REFERENCES
Andersson, Lars-Gunnar (1975), Form and Function of Subordinate
Clauses. Univendty of Goteborg, Dept. of Linguistics.
(Gothenburg Monographs in Linguistics I.)
Beckman, Natanae1 (1922) , En sko1grammatisk fraga i sprakhis­
torisk be1ysning. Skrifter utgivna av Modersma1s1ararnas
forening. Minnesskrift, 70-89.
Greenberg, Joseph (1963) , Some Universals of Grammar with Par­
ticular Reference to the Order of Meaningful Elements.
In Greenberg, J. (ed.) , Universals of Language. Cambridge,
Mass. and London.
Hammarberg, Bjorn (1975) , Svenska som ma1sprak. Sprakvard, Tid­
skrift utgiven av Svenska spraknamnden 2/1975, 12-15.
Hammarberg, Bjorn and Viberg, Ake (1975), Platshal1artvanget,
ett syntaktiskt problem i svenskan for invandrareo SSM
Report 2, University of Stockholm, Institute of Linguis­
tics.
Jespersen, Otto (1924) , The Philosophy of Grammar. London.
Kuno, Susumu (1974) , The Position of Relative Clauses and Con­
junctions. Linguistic Inquiry V, 117-136.
Lambton, Ann K.S. (1974) , Persian Grammar. Cambridge.
Lindberg, Ebba (1973) , Studentsvenska. Studier i stockho1ms­
abiturienters examensuppsatser 1864-1965'''' Stockholm.
Te1eman, U1f (1967) , Bisatser i ta1ad svenska, in Holm, Gosta
(ed.) , Svenskt ta1sprak. Stockholm.
Thorell, 010f (1973) , Svensk grammatik. Stockholm.